Observing What the Timelapse Captures
A Miri Timelapse incubator continuously photographs developing embryos over several days. Today, reviewing those images and assessing developmental stages is a manual process performed by trained embryologists.
We are prototyping a system that monitors the timelapse feed locally and tracks embryo stages as they progress — identifying transitions and flagging changes without requiring constant human attention.
How It Works
The model runs on the clinic's local hardware, alongside the MiriConnect server. It processes timelapse frames as they arrive, classifying the current developmental stage and detecting when an embryo moves from one stage to the next.
Privacy is a core constraint. By running inference locally, no embryo imagery leaves the clinic network. The system observes and reports — it does not transmit raw data externally.
Where We Are
This is early-stage work. The prototype is functional but not yet validated for clinical use. The current focus is on training accuracy across different imaging conditions and incubator configurations.
The Longer View
If the model proves reliable, it becomes a layer within MiriConnect that gives embryologists a structured, time-stamped record of each embryo's development — reducing the chance of missed transitions and supporting better-informed decisions during the treatment cycle.